Product manager interviews probe product thinking, comfort with metrics, and how you decide what not to build. Below are the most common questions with model answers.
Every question comes with a model answer you can compare yours against.
RICE scores each item by Reach times Impact times Confidence divided by Effort, which forces the estimates to be explicit and comparable. The score is not the decision — its value is that it exposes where two people disagree, usually on impact or confidence. Frameworks also need a strategic filter on top, because the highest-scoring item is sometimes irrelevant to this quarter's goal.
A North Star metric is the single measure that best captures the value customers receive and that predicts sustainable growth. For a marketplace it is usually completed transactions or gross merchandise value, not registrations, because a registration with no transaction created no value for anyone. A good North Star is hard to game — a metric like page views rewards behaviour that damages the product.
A usable hypothesis states the belief, the change, the expected effect, and the measurement: we believe that shortening onboarding to three steps will increase week-one activation from 22 to 28 percent, and we will know within two weeks. Naming the number in advance is what stops the team from declaring victory after the fact. It should also state what result would make you abandon the idea.
You look at adoption first — whether anyone found and used it — then retention of that usage, then the impact on the outcome metric the feature was meant to move. A feature can be used and still be a failure if it cannibalises a better flow or raises support volume. The discipline that makes this honest is defining the success threshold before launch rather than choosing a metric that looks good afterwards.
Treat it as information, not resistance: ask what specifically drives the estimate, then work on scope rather than pressure. Usually there is a smaller version that tests the same hypothesis, or a phased delivery where the risky part comes later. Pushing a team to commit to a date they do not believe produces quality debt that costs more than the delay would have.
You talk to users about their actual past behaviour rather than asking whether they would like a hypothetical feature, because people are unreliable predictors of their own future actions. Five to eight interviews per segment usually surfaces the dominant patterns, after which new interviews mostly repeat what you have heard. Discovery ends when you can predict the next interview's answers.
An MVP is the smallest thing that produces a real learning from real users about the riskiest assumption. The scoping question is not what can we cut but what must be true for this to work, and what is the cheapest way to test that. The common failure is shipping a small version of the full vision instead of a targeted test — you get a weak product and no learning.
Retention is the share of users still active after a period, churn is its inverse, and LTV is the total value a customer generates before leaving. They compound: a small retention improvement raises LTV disproportionately, which raises what you can afford to spend on acquisition. Improvement almost always starts at the beginning of the lifecycle — fixing week-one activation moves the whole curve more than any win-back campaign.
In a real interview you speak under pressure. Cam asks these same questions, scores every answer, and shows exactly what to fix.
Practice a Product Manager interview →